Development of a New Algorithm to Identify Clear Sky MSU Data Using AMSU-A Data for Verification

Observations from the microwave sounding unit (MSU), from 1978 to 2006 and its successor, the Advanced Microwave Sounding Unit-A (AMSU-A, 1998–present), and onboard the National Oceanic and Atmospheric Administration’s polar-orbiting satellites have been widely used for estimating global climate trends. The MSU has a long-term data set, but it is difficult to obtain cloud information like the cloud liquid water path (LWP) directly from this data set. To monitor and investigate the cloud effect on global upper air temperature trends using MSU observations, a cloud detection algorithm must be developed for the MSU. Considering the similar center frequencies of MSU channel 1 and AMSU-A channel 3, a new cloud detection algorithm is established based on the differences between observed and model-simulated brightness temperatures ( $O$ - $B$ - $\mu $ ( $\alpha$ )- $\mu $ ( $\phi $ )) of MSU channel 1 or AMSU-A channel 3 over oceans, where $\mu $ ( $\phi$ ) is a latitudinal dependent global mean bias and $\mu $ ( $\alpha$ ) is the global mean bias depending on scan angle. If a data point satisfies the condition of $O$ - $B$ - $\mu $ ( $\alpha$ )- $\mu (\phi) \ge 1$ K, it is removed from the clear sky data set. In order to ensure that those points that are partially affected by the clouds, such as clear and cloud mixed fields-of-views located near cloud edges and/or within optically thin clouds, all data points within the 60-km radial distance of the detected point are removed. Validated with the AMSU-A derived LWP retrievals, about 50% of the clear sky data are successively identified and 99% of the cloudy radiances are successively removed for all NOAA-15 AMSU-A data on January 15, April 15, July 15, and October 15, 2002.